Re: RED state exception (trap type 0x64) on U5 reboot

[Peter Hurley <peter@hurleysoftware.com>]

[+cc Christoph Lameter, Pekka Enberg ]

Hi Meelis,

On 06/16/2014 05:21 AM, Meelis Roos wrote:
> Back to an old dragon that seems to have more heads than thought.
> Background is that I got RED state exceptions from recursive faults on
> sparc64 during reboot via PROM, from running PROM code where PROM has
> all the cobtrol over the machine. In Sun Ultra 5 and Sun Blade 100 it
> resulted in RED state exception that looped and hung the machine. This
> is still a problem in current kernels.
>
> Now while debugging a different issue on Sun Fire V100, I noticed a
> different crash dump on reboot, showing also recursive fault from PROM
> space. The messages are different (because of newer PROM generation?)
> but the problem seems same, except no hang happens - that's why I had
> not noticed it before.
>
> Since V100 has LOM (remote lights-out management), I could test it more
> easily and decided to bisect it.

Is still only with the SLAB allocator and modular CONFIG_SUN_OPENPROMFS?

>>> but it came out clearly finally
>>> (each bad commit was clearly bad, each good commit was tested for 3
>>> reboots without a problem). Bisect resulted in his commit being at
>>> fault:
>>>
>>> 8cb06c983822103da1cfe57b9901e60a00e61f67 is the first bad commit
>>> commit 8cb06c983822103da1cfe57b9901e60a00e61f67
>>> Author: Peter Hurley<peter@hurleysoftware.com>
>>> Date:   Sat Jun 15 10:21:18 2013 -0400
>>>
>>>       n_tty: Remove alias ptrs in __receive_buf()
>>>
>>>       The char and flag buffer local alias pointers, p and f, are
>>>       unnecessary; remove them.
>>>
>>>       Signed-off-by: Peter Hurley<peter@hurleysoftware.com>
>>>       Signed-off-by: Greg Kroah-Hartman<gregkh@linuxfoundation.org>
>>>
>>> :040000 040000 ddc901fe810f43bc06a64397735b469b11e403e8
>>> 96d92e4e242c4b2ff11b25c005bccd093865b350 M  drivers
>
> And it was the same commit [8cb06c983822103da1cfe57b9901e60a00e61f67]
> there. So something seems to trigger with this commit.
>
> ttyS0 is sunsu conole on V100. Ultra 5 had sunsab. So the serial driver
> seems to be at least different.
>
> David, can you suggest a way to dump the whole state of sparc64 MMU to
> see if we leave some state different than before on reboot PROM call?
>